Back

Projections of wastewater as an indicator of COVID-19 cases in corrections facilities: a modelling study

Han, D.; Linares, P.; Holm, R. H.; Chandran, K.; Smith, T. R.

2023-10-31 public and global health
10.1101/2023.10.31.23296864 medRxiv
Show abstract

BackgroundAlthough prison facilities are not fully isolated from the communities they are located within, the majority of the population is confined and requires high levels of health vigilance and protection. This study sought to examine the dynamic relationship between facility level wastewater viral RNA concentration and probability of at least one positive COVID-19 case within the facility. MethodsThe study period was January 11, 2021 through May 12, 2023. Wastewater samples were collected and analyzed for SARS-CoV-2 (N1) and pepper mild mottle virus (PMMoV) three times per week across 14 prison facilities in Kentucky (USA). Confirmed positive clinical case reports were also provided. A hierarchical Bayesian spatial-temporal model with a latent lagged process was developed. FindingsWe modeled a facility-specific SARS-CoV-2 (N1) normalized by PMMoV wastewater ratio associated with at least one COVID-19 facility case with an 80% probability. The ratio differs among facilities. Across the 14 facilities, our model demonstrates an average capture rate of 94{middle dot}95% via the N1/PMMoV threshold with pts [≥] 0{middle dot}5. However, it is noteworthy as the pts threshold is set higher, such as at 0{middle dot}9 or above, the models average capture rate reduces to 60%. This robust performance underscores the models effectiveness in accurately detecting the presence of positive COVID-19 cases of incarcerated people. InterpretationThe findings of this study provide a correction facility-specific threshold model for public health response based on frequent wastewater surveillance.

Matching journals

The top 5 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.